I’m Giovanni Raimondo Quaratino, currently pursuing an MSc in Data Science and Management from LUISS Guido Carli, where I also earned my Bachelor’s degree in Political Science.
Dedicated data scientist, striving to make a meaningful impact in both business and academia.
Work Experience
European Central Bank
- Developed and deployed the architecture of a Variational Autoencoder (VAE) to perform anomaly detection on multivariate time series FINREP data using PyTorch, and Optuna.
- Designed and implemented a comprehensive Data Engineering pipeline, enabling efficient ETL processes with Apache Iceberg and PySpark, integrating MLflow for MLOps.
- Engineered, built and deployed a Streamlit application to automatically ingest, transform, and harmonize interest rate data from 20 euro area banks, leveraging Python-based ETL processes.
- Architected and deployed a Multi-Agent Retrieval Augmented Generation pipeline leveraging LangGraph, hybrid search with rank fusion, and contextual retrieval.
- Pretrained and fine-tuned an innovative version of the ModernBERT-base model for Monetary Policy Sentiment Analysis, achieving state-of-the-art performance.
- Researched and developed Temporal Kolmogorov-Arnold Networks (TKANs) for inflation forecasting, integrating textual data to enhance model performance and provide insights into economic trends.
02/2024 – Ongoing
Mashfrog Group
- Implemented Process Mining techniques using pm4py Python library and SAP Signavio, achieving a 15% improvement in process efficiency by identifying bottlenecks and streamlining workflows
- Utilized the “moto” Python library to effectively mock AWS Lambda and S3 services, enabling unit testing on data streaming processes
- Improved algorithms for train delay prediction by 30% using Machine Learning models
11/2022 – 06/2023
LUISS Data Lab
11/2022 – 01/2023
- Designed a Machine Learning pipeline using Amazon SageMaker to analyze data from social media using BERT- based state-of-the-art Natural Language Processing algorithms
- Performed network analysis to discover interactions among online toxic communities
American Embassy to Rome
- Developed an Artificial Intelligence based chatbot to guide American citizens through consular procedures
- Performed discrete event simulation in Python, reducing 4th of July celebration queue waiting time by 20%
05/2022 – 10/2022
